A Framework for Representing Interaction Tasks based on Tactile Data

This paper describes a framework for representing physical interaction tasks using tactile feedback. Although contact feedback has been widely exploited to control interaction with objects, the direct use of tactile information in designing and representing physical interaction rules has not received comparable attention in the literature. The missing link between algorithms implementing models of interaction and frameworks providing tactile information is one of the possible reasons. The major contribution of the paper is a working method to build a map of tactile sensors attached to a robot body and a control law using such a map to tune physical interaction with an external object. Experiments are used to validate the approach.